Introduce the 'completion_extra_word_chars' option

This string option is used to get all the additional characters
that should be considered as "word" character for the purpose of
insert mode completion.

Fixes #758

README.asciidoc

@@ -899,6 +899,9 @@ Some options are built in Kakoune, and can be used to control it's behaviour:
      where the completion apply in the buffer.
  * `static_words` _str-list_: list of words that are always added to completion
      candidates when completing words in insert mode.
+ * `completions_extra_word_chars` _str_: a string containing all additional character
+     that should be considered as word character for the purpose of insert mode
+     completion.
  * `autoreload` _enum(yes|no|ask)_: auto reload the buffers when an external
    modification is detected.
  * `debug` _flags(hooks|shell|profile)_: dump various debug information in

doc/manpages/options.asciidoc

@@ -124,9 +124,13 @@ Builtin options
 		to define where the completion apply in the buffer, and the
 		other strings are the candidates
 
-	*static_words* 'str-list':::
-		list of words that are always added to completion candidates
-		when completing words in insert mode
+*static_words* 'str-list'::
+	list of words that are always added to completion candidates
+	when completing words in insert mode
+
+*completions_extra_word_chars* 'str'::
+	a string containing all additional character that should be considered
+	as word character for the purpose of insert mode completion.
 
 *autoreload* 'enum(yes|no|ask)'::
 	auto reload the buffers when an external modification is detected

src/insert_completer.cc

@@ -76,12 +76,15 @@ WordDB& get_word_db(const Buffer& buffer)
 template<bool other_buffers>
 InsertCompletion complete_word(const SelectionList& sels, const OptionManager& options)
 {
+    StringView extra_word_char = options["completion_extra_word_char"].get<String>();
+    auto is_word_pred = [extra_word_char](Codepoint c) { return is_word(c) or contains(extra_word_char, c); };
+
     const Buffer& buffer = sels.buffer();
     ByteCoord cursor_pos = sels.main().cursor();
 
     using Utf8It = utf8::iterator<BufferIterator>;
     Utf8It pos{buffer.iterator_at(cursor_pos), buffer};
-    if (pos == buffer.begin() or not is_word(*(pos-1)))
+    if (pos == buffer.begin() or not is_word_pred(*(pos-1)))
         return {};
 
     ByteCoord word_begin;
@@ -92,7 +95,7 @@ InsertCompletion complete_word(const SelectionList& sels, const OptionManager& o
         Utf8It end{buffer.iterator_at(sels[i].cursor()), buffer};
         Utf8It begin = end-1;
         if (not skip_while_reverse(begin, buffer.begin(),
-                                   [](Codepoint c) { return is_word(c); }))
+                                   [&](Codepoint c) { return is_word_pred(c); }))
             ++begin;
 
         if (i == sels.main_index())
@@ -101,7 +104,7 @@ InsertCompletion complete_word(const SelectionList& sels, const OptionManager& o
             prefix = buffer.string(word_begin, end.base().coord());
         }
 
-        skip_while(end, buffer.end(), [](Codepoint c) { return is_word(c); });
+        skip_while(end, buffer.end(), [&](Codepoint c) { return is_word_pred(c); });
 
         auto word = buffer.string(begin.base().coord(), end.base().coord());
         ++sel_word_counts[word];

src/main.cc

@@ -266,6 +266,9 @@ void register_options()
                        "%val{bufname} %val{cursor_line}:%val{cursor_char_column} "_str);
     reg.declare_option("debug", "various debug flags", DebugFlags::None);
     reg.declare_option("readonly", "prevent buffers from being modified", false);
+    reg.declare_option("completion_extra_word_char",
+                       "Additional characters to be considered as words for insert completion",
+                       ""_str);
 }
 
 struct convert_to_client_mode

src/word_db.cc

@@ -10,7 +10,7 @@ namespace Kakoune
 
 using WordList = Vector<StringView>;
 
-static WordList get_words(StringView content)
+static WordList get_words(StringView content, StringView extra_word_chars)
 {
     WordList res;
     using Utf8It = utf8::iterator<const char*>;
@@ -19,7 +19,7 @@ static WordList get_words(StringView content)
     for (Utf8It it{word_start, content}, end{content.end(), content}; it != end; ++it)
     {
         Codepoint c = *it;
-        const bool word = is_word(c);
+        const bool word = is_word(c) or contains(extra_word_chars, c);
         if (not in_word and word)
         {
             word_start = it.base();
@@ -36,9 +36,14 @@ static WordList get_words(StringView content)
     return res;
 }
 
+static StringView get_extra_word_chars(const Buffer& buffer)
+{
+    return buffer.options()["completion_extra_word_char"].get<String>();
+}
+
 void WordDB::add_words(StringView line)
 {
-    for (auto& w : get_words(line))
+    for (auto& w : get_words(line, get_extra_word_chars(*m_buffer)))
     {
         auto it = m_words.find(w);
         if (it == m_words.end())
@@ -56,7 +61,7 @@ void WordDB::add_words(StringView line)
 
 void WordDB::remove_words(StringView line)
 {
-    for (auto& w : get_words(line))
+    for (auto& w : get_words(line, get_extra_word_chars(*m_buffer)))
     {
         auto it = m_words.find(w);
         kak_assert(it != m_words.end() and it->second.refcount > 0);
@@ -66,14 +71,44 @@ void WordDB::remove_words(StringView line)
 }
 
 WordDB::WordDB(const Buffer& buffer)
-    : m_buffer{&buffer}, m_timestamp{buffer.timestamp()}
+    : m_buffer{&buffer}
 {
+    buffer.options().register_watcher(*this);
+    rebuild_db();
+}
+
+WordDB::WordDB(WordDB&& other)
+    : m_buffer{std::move(other.m_buffer)},
+      m_lines{std::move(other.m_lines)},
+      m_words{std::move(other.m_words)},
+      m_timestamp{other.m_timestamp}
+{
+    kak_assert(m_buffer);
+    m_buffer->options().unregister_watcher(other);
+    other.m_buffer = nullptr;
+
+    m_buffer->options().register_watcher(*this);
+}
+
+WordDB::~WordDB()
+{
+    if (m_buffer)
+        m_buffer->options().unregister_watcher(*this);
+}
+
+void WordDB::rebuild_db()
+{
+    auto& buffer = *m_buffer;
+
+    m_words.clear();
+    m_lines.clear();
     m_lines.reserve((int)buffer.line_count());
     for (auto line = 0_line, end = buffer.line_count(); line < end; ++line)
     {
         m_lines.push_back(buffer.line_storage(line));
         add_words(m_lines.back()->strview());
     }
+    m_timestamp = buffer.timestamp();
 }
 
 void WordDB::update_db()
@@ -118,6 +153,12 @@ void WordDB::update_db()
     m_lines = std::move(new_lines);
 }
 
+void WordDB::on_option_changed(const Option& option)
+{
+    if (option.name() == "completion_extra_word_char")
+        rebuild_db();
+}
+
 int WordDB::get_word_occurences(StringView word) const
 {
     auto it = m_words.find(word);

src/word_db.hh

@@ -13,12 +13,13 @@ namespace Kakoune
 using RankedMatchList = Vector<RankedMatch>;
 
 // maintain a database of words available in a buffer
-class WordDB
+class WordDB : public OptionManagerWatcher
 {
 public:
     WordDB(const Buffer& buffer);
+    ~WordDB();
     WordDB(const WordDB&) = delete;
-    WordDB(WordDB&&) = default;
+    WordDB(WordDB&&);
 
     RankedMatchList find_matching(StringView str);
 
@@ -28,6 +29,10 @@ private:
     void add_words(StringView line);
     void remove_words(StringView line);
 
+    void rebuild_db();
+
+    void on_option_changed(const Option& option) override;
+
     struct WordInfo
     {
         StringDataPtr word;